Routing

 View Only

IMPORTANT MODERATION NOTICE

This community is currently under full moderation, meaning  all posts will be reviewed before appearing in the community. Please expect a brief delay—there is no need to post multiple times. If your post is rejected, you'll receive an email outlining the reason(s). We've implemented full moderation to control spam. Thank you for your patience and participation.



Monitoring PFE memory usage

  • 1.  Monitoring PFE memory usage

    This message was posted by a user wishing to remain anonymous
    Posted 12 days ago
    This message was posted by a user wishing to remain anonymous

    Hello,

    we're using an MX240 with 2x RE-S-2000 running as BGP edge router

    We've noticed issues with packet forwarding, and noticed the log messages complaining about "no more memory":

    fpc1 RT: Failed prefix add IPv4 - 64.188.46/24 (No memory) on FE 0
    fpc1 RT: Failed prefix add IPv4 - 64.188.46/24 (No memory) on FE 0
    fpc1 RT: IPv4:0 - 64.188.46/24 (add rt entry into jtree failed)
    re1 eventd: sendto: No route to host
    fpc1 RT: Failed prefix add IPv6 - 2401:7540::/32 (No memory) on FE 0
    fpc1 NH: unable to create RNH_HOLD ENH entry
    fpc1 RT-HAL,rt_msg_handler,673: route check failed 7

    Now while we reduced the number of routes and do not see the messages any longer (after chassis reboot), I'm wondering what our "margin of error" is, exactly.

    The messages are referenced in KB24391, for example:

    Juniper Networks


    But while the article says "The router's PFE reached its limit and cannot add any more routes.", there is no definite hint which "memory", exactly, is meant here and which show command (or SNMP OID) I would use to see the exact current utilzation to verify that there's enough room now.

    show chassis routing-engine shows
    RE0:
    DRAM 3584 MB (4096 MB installed)
    Memory utilization 33 percent

    RE1:
    DRAM 3584 MB (4096 MB installed)
    Memory utilization 12 percent

    (3584 out of 4096 MB certainly is neither 33 nor 12 percent - likely bloated by buffer/cache usage? Does the MB valur have any relevance?)

    AND

    show system resource-monitor fpc
    FPC Resource Usage Summary

    Free Heap Mem Watermark : 20 %
    Free NH Mem Watermark : 20 %
    Free Filter Mem Watermark : 20 %

    Heap ENCAP mem NH mem FW mem
    Slot # % Free PFE # % Free % Free % Free
    1 63 0 99 36 72
    1 99 36 72
    2 99 36 72
    3 99 36 72
    2 62 0 99 36 72
    1 99 36 72
    2 99 36 72
    3 99 36 72

    -> which suggests 36% free Next-Hop memory at the  moment.

    The article

    https://www.juniper.net/documentation/us/en/software/junos/network-mgmt/topics/topic-map/troubleshooting-system-performance-with-resource-monitoring-methodology.html

    seems to show SNMP OIDs that I could use, although snmpwalk does not output those OIDs at the moment (maybe needs manual configuration?)


    I do not have the values for the situation where the NH "no memory" messages occurred to verify if any of those would've displayed 100% / 0%. Am I looking at the right values?

    Or is there another place where PFE memory utilization can be checked?